Build OpenCV with CUDA

Build OpenCV with CUDA

#opencv   #build   #cuda  

---

Table of Content

Build OpenCV on x64 ubuntu machine

Prerequisites

• install nvidia driver …
• install cudnn …
• install cmake gui
• install all opencv dependencies

sudo apt install cmake-qt-gui

install dependencies

sudo apt-get update
sudo apt-get upgrade
sudo apt-get install build-essential cmake unzip pkg-config
sudo apt-get install libjpeg-dev libpng-dev libtiff-dev
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev
sudo apt-get install libv4l-dev libxvidcore-dev libx264-dev
sudo apt-get install libgtk-3-dev
sudo apt-get install libblas-dev liblapack-dev gfortran
sudo apt-get install python3-dev

Download source

wget https://github.com/opencv/opencv/archive/4.5.5.zip
wget https://github.com/opencv/opencv_contrib/archive/4.5.5.zip

# unzip

---

build process

• set all option
• run configure
• run generate
• run make -j 8 from/on build folder
• check

config

• Select source and build folder

• config python

• config cuda

• gpu arch wikipedia cuda gpu support

• dnn

• blas

• fast math

• contrib

• pkgconfig

config result

• show my changes

Commandline options:
-DOPENCV_EXTRA_MODULES_PATH:PATH="/home/user/opencv/opencv_contrib-4.5.5/modules" \
-DOPENCV_GENERATE_PKGCONFIG:BOOL="1" \
-DWITH_CUDA:BOOL="1" \
-DCUDA_ARCH_BIN:STRING="7.5" \
-DOPENCV_DNN_CUDA:BOOL="1" \
-DENABLE_FAST_MATH:BOOL="1" \
-DPYTHON3_EXECUTABLE:FILEPATH="/usr/bin/python3" -DBUILD_opencv_world:BOOL="1" \
-DCUDA_FAST_MATH:BOOL="1" 


Cache file:
OPENCV_EXTRA_MODULES_PATH:PATH=/home/user/opencv/opencv_contrib-4.5.5/modules
OPENCV_GENERATE_PKGCONFIG:BOOL=1
WITH_CUDA:BOOL=1
CUDA_ARCH_BIN:STRING=7.5
OPENCV_DNN_CUDA:BOOL=1
ENABLE_FAST_MATH:BOOL=1
PYTHON3_EXECUTABLE:FILEPATH=/usr/bin/python3
BUILD_opencv_world:BOOL=1
CUDA_FAST_MATH:BOOL=1

---

Check

• create new virtual env and activate
• run setup.py from build/python_loader folder

import cv2
cv2.__version__
cv2.cuda.getCudaEnabledDeviceCount()